Data-Driven Vulnerable Community Identification During Compound Disasters

This paper aims to introduce research directions to identify vulnerable communities during compound disasters, such as the intersection of hurricanes and pandemics like COVID-19. We suggest integrating mobility, socioeconomic, and geographic factors to develop a comprehensive vulnerability index. By analyzing mobility data, U.S. Census socioeconomic data, disaster data, pandemic data, and geographic data, we extract information like hospital visits, business closures, increased travel distances, decreased inbound movement to commercial census block groups (CBGs), decreased outbound from home CBGs, and changes in mobility trends to various categories of stores, to develop a vulnerability index that integrates all of these factors. This index aims to capture and analyze the unique impacts of natural disasters on communities, especially those exacerbated by simultaneous disaster scenarios. We also propose a model design that can capture the spatial and temporal nature of the data and can be trained to perform the dual task of vulnerability index calculation in a disaster scenario along with predicting pandemic growth features in case of a compound disaster.